Jezero crater, Mars: application of the deep learning NOAH-H terrain classification system

We applied a deep learning terrain classification system, the ‘Novelty or Anomaly Hunter – HiRISE’ (NOAH-H), originally developed for ExoMars landing sites in Oxia Planum and Mawrth Vallis, to Mars 2020 Perseverance rover site Jezero crater. NOAH-H successfully classified four HiRISE images of even though landforms study area were slightly different from those training dataset. mosaicked rasters compared them with manually generated photogeological map, Ingenuity helicopter images. find that grouped classes correspond well humanmade map individual are corroborated by available ground-truth conclude our products can be refined feeding into traversability analysis Rosalind Franklin at they also used aid mapping process.